Three-dimensional (3D) models of a building may be generated based on two-dimensional (2D) digital images taken of the building. The digital images may be taken via aerial imagery, specialized camera-equipped vehicles, or by a user with a camera. The 3D building model is a digital representation of the physical, real-world building. An accurate 3D model may be used to derive various building measurements or estimate design and renovation costs.
However, not all 2D images can produce an accurate 3D model. For example, if the 2D images used to create a 3D model are only of the front of a building, the back of the 3D model of the building is likely to be inaccurate. Similarly, if portions of a building are obscured by other objects in the image, then the obscured portions may not be represented accurately in the 3D model. As another example, if the image capture device was tilted while capturing the photo, the proportions of building features derived from the image may be skewed.
Generating an accurate 3D model from such images may be difficult, time-consuming, or impossible, depending on the quality of the images. In this context, a set of photos is deemed to have “high quality” if deriving an accurate 3D model from the set of photos is possible without extraordinary effort, while a set of photos is deemed to have “low quality” if deriving an accurate 3D model from the set of photos is relatively difficult or impossible. Thus, in this context “quality” does not relate to the artistic or aesthetic value of a photo.
In situations where individuals must travel to the site of a building, take photos of it, and then submit the photos to a 3D modeling service, an enormous amount of resources can be wasted if the photos turn out to be of low quality. In such situations, the amount of effort required to generate an accurate 3D model from the photos may be impractically high. Further, in such cases it may even be necessary to send someone back out to the building site to take a second round of photos. Clearly, it is desirable to avoid such wasted efforts.
The approaches described in this section are approaches that could be pursued, but not necessarily approaches that have been previously conceived or pursued. Therefore, unless otherwise indicated, it should not be assumed that any of the approaches described in this section qualify as prior art merely by virtue of their inclusion in this section.